FPGA Realization Of Temperature Compensation Of Pressure Sensor Based On BP Neural Network

With the development of The Times,pressure sensors are playing a more and more important role in our life,and silicon piezoresistive pressure sensors are more widely used.However,there is always the problem of temperature drift in its application,so many temperature compensation methods have emerged.In recent years,neural network has become an important way of temperature compensation for pressure sensors.However,most people only realized the temperature compensation of neural network for pressure sensors on Matlab software and single-chip microcomputer,without giving play to the parallel characteristics of neural network,and the compensation speed is slow,so it is difficult to realize real-time compensation.This design is based on BP neural network of FPGA for temperature compensation of silicon piezoresistive pressure sensor.The BP neural network was optimized by genetic algorithm,the optimal weight and threshold were obtained by Matlab,and the temperature compensation model of pressure sensor based on GA-BP neural network was established.The problem of difficult hardware implementation of Tanh activation function was solved by polynomial piecewise fitting.The BP neural network temperature compensation model was realized by FPGA,simulated by Modelsim software,downloaded to FPGA development board,and tested online by Signal Tap II.The experimental results show that the sensitivity temperature drift and zero temperature drift of the pressure sensor after temperature compensation are reduced by two orders of magnitude,and the compensation speed is greatly improved,which can realize real-time compensation theoretically.